A prion is an infectious protein, a concept derived from studies of mammalian spongiform encephalopathies. We discovered that the yeast Saccharomyces cerevisiae can be infected with two prions, two non-chromosomal genetic elements whose properties are those predicted for an infectious protein, not a nucleic acid replicon or a virus. The first, [URE3] is an altered form of Ure2p, the protein product of the chromosomal URE2 gene important in regulation of nitrogen metabolism. The second, [PSI], is an altered form of Sup35p, a subunit of the translation release factor and product of the chromosomal SUP35 gene. We found that Ure2p is more resistant to protease digestion in [URE3] strains than in wild-type strains, supporting the prion model for [URE3]. This protease resistance is not a constant concomitant of derepression of nitrogen metabolism. The N-terminal 65 aminoacid residues of Ure2p is sufficient to propagate [URE3], in the complete absence of the C-terminal nitrogen regulation domain. The overexpression of the URE2 protein, and not the URE2 RNA, is what induces the de novo formation of [URE3]. The C-terminal nitrogen regulation domain is only inactivated when covalently attached to the N-terminal domain, showing that [URE3] is propagated by interactions between the N-terminal prion domains. Certain deletions within the C-terminal nitrogen regulation domain of Ure2p result in a 100-fold increase the frequency with which the [URE3] prion arises. In fact, either of two non-overlapping parts of Ure2p can, when overexpressed, induce the de novo appearance of [URE3]. We have now shown that Ure2p is aggregated in vivo in cells carrying the prion [URE3]. This aggregation requires the prion domain, and disappears in cells cured of [URE3]. We find that the Ure2p prion domain (residues 1-65) forms amyloid filaments in vitro. Other similar size fragments of Ure2p that do not include the prion domain do not form amyloid. The prion domain promotes the formation of amyloid filaments by the purified native soluble Ure2p. This reaction is specific in that the Ure2p prion domain does not promote amyloid formation by other proteins, nor does the amyloid-forming peptide Abeta promote amyloid formation by Ure2p. The properties of Ure2p amyloid formation in vitro reflect and explain the prion properties of [URE3] in vivo. We propose that the [URE3] prion is an infectious amyloidosis. We find that overexpression of fragments of Ure2p, or expression of certain Ure2p-GFP fusion proteins at normal levels results in efficient curing of the [URE3] prion. This phenomenon may be due to interruption of the growth of the amyloid 'crystals' due to the fragments or fusion proteins, and suggests a new approach to the treatment of amyloid diseases. We find that the Mks1 protein is essential for the de novo formation of the [URE3] prion. Mks1 activity is negatively regulated by the Ras - cAMP pathway, and we find that activation of Ras2p prevents de novo [URE3] prion formation by inactivating Mks1. We find that the Hsp104 chaperone is necessary for [URE3] prion propagation. We also find that overexpression of the Hsp40-family chaperone Ydj1p can cure the [URE3] prion.